Short, substantive notes from inside the engineering work. Written in the company voice rather than bylined; published when the technical decision is settled enough to explain publicly without compromising IP or competitive position.
2026 · Q2
Geometry
~4 min read
Why three optical channels, not four or five.
The most common question we get from optics engineers reviewing the architecture is why we settled on three channels for full-arch dual-arch coverage rather than four, five, or more. The answer is a constraint cascade, not a free choice.
Each additional optical channel introduces a calibration interdependency that compounds across the manufacturing process. Two channels need a single relative-pose calibration. Three need three pairwise calibrations and a global consistency check. Four need six pairwise calibrations. Five need ten. The metrology cost grows quadratically while the field-of-view coverage gain grows sub-linearly.
The horseshoe geometry of a typical adult dental arch can be covered, with overlap margin, by three channels positioned at the anterior labial midline and the two posterior buccal regions. The anterior channel sees both lower incisors and upper incisors simultaneously. The posterior channels see the molar and premolar regions on each side. Coverage at the canine-premolar transition is provided by overlap from the anterior channel on the lingual side and the buccal channels on the buccal side.
Adding a fourth channel — for example, a dedicated lingual channel for the molar regions — gains marginal coverage at the cost of doubling the calibration matrix. The marginal coverage gain is in surface regions that are already imaged at acceptable density; the calibration cost is borne on every unit shipped.
The three-channel architecture is not the maximum possible coverage architecture. It is the maximum-coverage architecture that does not pay a manufacturing-cost penalty disproportionate to the imaging-quality gain. The decision was driven by metrology economics, not by the optical limit.
2026 · Q2
Calibration
~5 min read
Factory calibration vs. in-field calibration: the architectural choice.
A dual-arch capture architecture has two reasonable approaches to maintaining inter-channel pose accuracy: factory-set with mechanical retention, or in-field re-calibration with a reference target. We chose the first. Here is the reasoning.
In-field calibration would let the system tolerate gradual mechanical drift over the device lifetime — a real concern for any structural component subjected to thousands of autoclave cycles, occlusal bite loads, and routine handling. The trade-off is that the clinician (or DSO clinical staff member) becomes responsible for periodic calibration runs, which introduces operator burden and creates a state-dependency in capture quality that we explicitly designed out of the architecture.
Factory calibration with mechanical retention pushes the engineering burden upstream. The shell material, internal mounting structure, and thermal-spreader layer are all selected and arranged to maintain optical-channel relative pose across the full expected service envelope: occlusal bite cycles, autoclave thermal cycles, drop-event mechanical shocks, and ambient temperature variation in operatory environments. The calibration retention claim is then validated as a bench test (Test 4.5 in the testing matrix) at 0, 25, 50, and 100+ autoclave cycles — a single validated claim made once at design release rather than a recurring per-unit operational requirement.
The architectural consequence is that material selection (shell polymer grade, mounting structure rigidity) and assembly tolerances (channel-to-substrate bond integrity) become part of the regulatory submission. Calibration is a documented design output, not a use-time procedure. This is the pattern most cleared intraoral scanners follow at the channel level; we extend it to the inter-channel level.
The decision is not absolute. A future generation of the platform could move to in-field calibration if the use-case calls for it — for example, a portable variant where the cost-benefit calculus changes. The Generation 1 architecture, however, treats calibration as a manufactured-in property of the device, not a runtime concern.
2026 · Q2
Materials
~4 min read
Polycarbonate, PEI, and the autoclave-cycle question.
The shell material decision sits at the intersection of biocompatibility, optical translucency, mechanical rigidity, and autoclave thermal cycling. Two material classes meet all four constraints — medical-grade polycarbonate and polyetherimide (PEI). The choice between them is more nuanced than a spec table suggests.
Polycarbonate is the lower-cost choice with established medical-device pedigree, broad supplier base, good optical clarity, and adequate autoclave tolerance for the service life we target. Its weak point is repeated steam exposure: hydrolysis of the carbonate linkage at elevated temperature progressively reduces mechanical properties over many cycles. For a device claiming ≥100 autoclave cycles, polycarbonate is on the edge of its envelope — tractable, but requires careful grade selection and stabilizer chemistry, and the cycle-life claim has to be validated rather than assumed.
PEI is the more expensive choice with a smaller medical-device supplier base but a substantially better autoclave envelope. Its glass transition temperature is well above the autoclave cycle peak, hydrolysis resistance is markedly better than polycarbonate, and dimensional stability across thermal cycles supports the calibration retention claim discussed above. The trade-offs are unit cost (notable on a high-volume disposable-adjacent device), optical clarity (slightly inferior to optical-grade polycarbonate but acceptable for our internal-channel application), and supplier diversity in medical grades.
The current design holds the choice open between the two grades pending the cycle-life validation results. The bench testing protocol exposes representative shell samples to the full claimed cycle count under realistic loading and humidity profiles, with mechanical and optical property measurements at intervals. The grade that passes with the most retained margin wins.
What the public site cannot show is that this decision is being made on data, not on materials science folklore. The diligence package documents the test protocol, sample plan, and decision criteria. What the site can show is that the decision is real and the criteria are written down.
Engineering notes are published when a technical decision is settled enough to explain without compromising IP or competitive position. Detailed validation data and decision criteria are documented in the diligence package available under NDA.
